license: apache-2.0
base_model: Helmcode/Qwen3.8-27B-Abliterated
library_name: gguf
tags:
- abliterated
- uncensored
- gguf
- reasoning
- multimodal
Qwen3.8-27B Abliterated — GGUF
GGUF quants of Helmcode/Qwen3.8-27B-Abliterated.
See that card for the abliteration method, honest measurements and the
safety note: this model complies with requests the base model refuses.
You are responsible for how you use it.
| file | size | RAM (approx) |
|---|---|---|
Q4_K_M |
16.7 GB | ~20 GB |
Q5_K_M |
19.3 GB | ~24 GB |
Q6_K |
22.1 GB | ~27 GB |
Q8_0 |
28.6 GB | ~34 GB |
Usage
llama-server --model qwen3.8-27b-abliterated-v2-Q4_K_M.gguf --jinja --port 8080
--jinja enables the native chat template with reasoning. Thinking is ON
by default and reasoning can be long — give generous max_tokens (>= 1000)
or disable thinking in the request.
License
Apache 2.0, inheriting the base model's license.